Neuroleptic-induced tardive dyskinesia (TD) is a serious public health problem among chronically mentally ill older patients. There have been very few prospective, long-term studies of the incidence of and risk factors for TD in this population. We propose to evaluate, over a five-year period, 700 psychiatric patients over age 50. At study entry, these patients will have had less than 1 month of total lifetime neuroleptic exposure, and will be neuroleptic-free for at least a month prior to study entry. The initial assessments will include psychiatric and neurologic examinations (with suitable rating scales), a through review of past medical and medication records, a specific clinical evaluation of motor function and of neurologic "soft" signs, a selected neuropsychological test battery, and an instrumental assessment of orofacial and limb motor function (including parkinsonism, involuntary movements and voluntary motor control). Patients will be assigned randomly to either haloperidol or thioridazine, and we will attempt to maintain them on the same neuroleptic throughout the study. The treatment will otherwise be individualized, with the goal always being to treat the patients with the lowest effective dose. Patients will be reexamined one month after initial assessment and that at 3-month intervals for evidence of TD. The neuropsychological and instrumental assessments will be repeated annually. All these evaluations will be done "blind" with respect to the other data (especially treatment). The main goals of our study are: 1) to estimate the incidence of TD in this older patient population, 2) to determine the risk factors for occurrence and precipitation of TD, as well as development of a so-called "malignant" form of TD, and 3) to determine the risk factors for persistence and severity of TD. Statistical methods will include survival analysis with covariates, and stepwise regression analysis. We will perform a quantitative neuropathologic study of the brains of the patients who die. The main strengths of the proposed work are: a large sample size, use of a neuropsychologic test battery, instrumental assessment of orofacial/limb motor function, a comparison of relative risk of TD with two most commonly used neuroleptics, and a neuropathologic study of the brains of TD patients.